2021-11-13

分析宏基因组往往是一件很漫长的流程，大神Pierre MARTIN编写了[metagwgs](https://forgemia.inra.fr/genotoul-bioinfo/metagwgs)工作流程可以使用集群对metagenomic data（Illumina HiSeq3000 or NovaSeq, paired, 2*150bp）进行分析，该流程涵盖了宏基因组分析所有必要步骤，还包含assembly, taxonomic annotation以及functional annotation of predicted genes。

输入raw data格式为.fastq或者.fastq.gz,该工作流程一共包含了8个步骤：

• 01_clean_qc (can ke skipped)

  • trims 接头序列和 deletes低质量的reads，相关软件： (Cutadapt, Sickle)
  • 抑制宿主基因干扰，相关软件： (BWA + Samtools + Bedtools)
  • 质量控制，相关软件： (FastQC)
  • cleaned reads的系统分类，相关软件： (Kaiju MEM + kronaTools + Generate_barplot_kaiju.py + merge_kaiju_results.py)

• 02_assembly

  • 从01_clean_qc 开始组装 (cleaned reads )或从raw reads (需要使用 --skip_01_clean_qc 参数) ，相关软件：(metaSPAdes or Megahit)
  • 评估组装后reads质量，相关软件： (metaQUAST)
  • 去重，相关软件： (BWA + Samtools + Bedtools)

• 03_filtering (can be skipped)

  • filters contigs with low CPM value (Filter_contig_per_cpm.py + metaQUAST)

• 04_structural_annot

  • makes a structural annotation of genes (Prokka + Rename_contigs_and_genes.py)

• 05_alignment

  • aligns reads to the contigs (BWA + Samtools)
  • aligns the protein sequence of genes against a protein database (DIAMOND)

• 06_func_annot

  • makes a sample and global clustering of genes (cd-hit-est + cd_hit_produce_table_clstr.py)
  • quantifies reads that align with the genes (featureCounts + Quantification_clusters.py)
  • makes a functional annotation of genes and a quantification of reads by function (eggNOG-mapper + best_bitscore_diamond.py + merge_abundance_and_functional_annotations.py + quantification_by_functional_annotation.py)

• 07_taxo_affi

  • taxonomically affiliates the genes (Samtools + aln2taxaffi.py)
  • taxonomically affiliates the contigs (Samtools + aln2taxaffi.py)
  • counts the number of reads and contigs, for each taxonomic affiliation, per taxonomic level (Samtools + merge_contig_quantif_perlineage.py + quantification_by_contig_lineage.py)

• 08_binning from nf-core/mag 1.0.0

  • makes binning of contigs (MetaBAT2)
  • assesses bins (BUSCO + metaQUAST + summary_busco.py and combine_tables.py from nf-core/mag)
  • taxonomically affiliates the bins (BAT)

A report html file is generated at the end of the workflow with MultiQC.

The pipeline is built using Nextflow, a bioinformatics workflow tool to run tasks across multiple compute infrastructures in a very portable manner.

Three Singularity containers are available making installation trivial and results highly reproducible.